GPU RMA testing

[benjii]

I sent my 290 back to OCUK due to the sudden appearance of artifacts in-game (when overclocked) and on the deskstop (at stock) while running a LTC miner.

I was just curious, how do they test them? Do they have a diagnostics program or is it just trying to replicate the issue with the same tools as everyone else (Valley, Afterburner etc)?

Also, does anyone else get nervous when sending items off for an RMA and think that they won't find the problem?

 

[Gregster]

I believe they run Heaven on a loop and something else to verify if it is broken.

 

[setter]

I do indeed get a bit nervous when rma'ing parts. Recently bought a 2400mhz patriot ram kit. It wouldnt run it's rated speed/timings using xmp or manual settings. Bsods galore and failed memtest but ran ok at 1600mhz. Sent it back anyway and managed to get it swapped for the kit in sig, (better timings too) which worked great using xmp straight away.

 

[Marine-RX179]

If you are getting artifact when overclocking the 290, it most likely just means you are not giving it enough voltage. I get artifact too when overclocking, but increasing the voltage or lower the core clock the artifact will go away.

What kinda of clock you were trying to push on the core with what voltage by the way? For my 290x I need like 1.412mV (+152mV) on 1200MHz and 1.350mV (+100mV) on 1170MHz to not get artifact...

 

[benjii]

If you are getting artifact when overclocking the 290, it most likely just means you are not giving it enough voltage. I get artifact too when overclocking, but increasing the voltage or lower the core clock the artifact will go away.

What kinda of clock you were trying to push on the core with what voltage by the way? For my 290x I need like 1.412mV (+152mV) on 1200MHz and 1.350mV (+100mV) on 1170MHz to not get artifact...

It wasn't a lack of voltage, it was rock solid at 1125mhz +50mv for days. Then suddenly I couldn't even do 1050mhz at +50mv without near instant artifacts leading to a crash in all 3d applications and the GPU and both VRMs were <60c.

 

[Marine-RX179]

It wasn't a lack of voltage, it was rock solid at 1125mhz +50mv for days. Then suddenly I couldn't even do 1050mhz at +50mv without near instant artifacts leading to a crash in all 3d applications and the GPU and both VRMs were <60c.

Do you have the same problem with the card on stock clock though? Cause if it doesn't and you rma it, you'd gonna get a "no fault found" and get charged...I think £25 as testing fee? If you are really not happy with the card, then return in under DSR; however if the DSR period has already passed, and the card doesn't have problem on stock clock, then I'm afraid you are stuck with it.

All cards and GPUs mileage varies...may be you should just try increasing the voltage and see if it would solve the artifact issue?

Also the real-world performance difference between 1150MHz and 1200MHz would be little anyway. I did the Sleeping Dogs bench at both 1150 and 1200, and there was only about 1-2fps difference on min, average and max frame rate. I myself sometimes wonder if I should leave the card on 1100~1125MHz so it would generate less heat and has lower power consumption...